Use of AI and imaging techniques to improve fisheries sustainability
Global fisheries supply 80 million tonnes of food for humans each year. Understanding the amount of catch removed from the sea, including the part of the catch that is returned alive or dead, is critical to informing good fisheries management and science. While it is a simple task to know how much catch is landed by a vessel, the amount of bycatch caught is often unknown and has to be estimated. This PhD is focused on using computer vision and AI to automate the task of collecting and processing data on bycatch composition in fisheries at sea using imaging techniques. To date, information on bycatch is usually recorded by observers on board vessels, a process that is time consuming, expensive, and achieves minimal coverage in space and time. Automating the data gathering process would create a step change in our ability to collect data. The project will provide data into the fishery improvement project 'Project UK'.
The objective of the PhD is to develop the necessary techniques to process and analyse images obtained from an HD camera and a 3D laser scanner. The challenge is to use machine learning to train a model to recognize the catch (scallops) and the bycatch (other species) and to differentiate these images from inert material (e.g. rocks). The reason for using the two approaches (HD camera and laser) is to compare results derived from each or combine the two sources of data in a multimodal system. The lead supervisor has worked previously with Bangor University and Aberystwyth University to make advances on the camera imaging aspect of the project. This is now at a mature stage with new hardware (i.e. a camera with on-board computing capability). The current state of workstream development means that algorithms have been developed such that video images can be sliced into still frame images of valid records ready for species and size determination. The laser aspect of the project is at a concept phase, but the technique is currently applied at Ulster University to scan seabed morphology and habitats, and hence it is known that the application would work in the proposed context.
The multidisciplinary nature of the project requires a large supervisory team. It is expected that the student will be based primarily at Heriot-Watt University but you will be expected to spend several months working with the teams at Aberystwyth University and Ulster University. This is an exciting project at the cutting edge of the use of technology to improve the sustainability of fisheries, with exceptional prospects for employment in academia, Government agencies, commercial fisheries, food processing, defense and other tech sectors.
The supervisory team brings the following skill sets to the project:
Prof Michel Kaiser (HWU): Fisheries data requirements to underpin sustainable fisheries, fishing industry and UK Government and agency linkages, direct engagement with fishing industry.
Dr Marta Vallejo (HWU): Machine learning and deep learning technologies for computer vision problems
Dr Chris McGonigle (UU): Marine surveying with optical and acoustic sensors applied to ecological monitoring, direct engagement with fishing industry and policy leads in NI and RoI.
Dr Bernie Tiddeman (AU): Computer vision and image processing with applications in biology, psychology, heritage and medicine.
Dr Marie Neal (AU): Computer Vision and image processing with applications in plant and marine biology.
Dr Natalie Hold (BU): Sustainable fisheries data collection and analysis, Welsh fishing industry and government linkages, PI on current crustacean imaging project, direct engagement with Welsh fishing industry.
The project is fully funded for 3.5 years and covers the PhD fees (UK fees) and stipend (currently £15 285 per annum) and has a generous travel and equipment budget.
This project is available to home (UK) or other students. The successful candidate will have a B.Sc. (2/1 or higher) and M.Sc. (distinction) or equivalent, and ideally additional experience in computer science, modelling, image analysis or engineering, ideally with some experience of applying these skills to other disciplines (e.g. in the life sciences). You will have good programming skills, preferably in Python or other advanced programming languages. Knowledge of Tensorflow, Keras, Pytorch and/or other deep learning frameworks would be advantageous. You will be highly self-motivated and confident enough to seek out solutions beyond the current team if required. Candidates that have some appreciation of fisheries or practical experience of working at sea would be preferred. The PhD will require sea-time on board fishing vessels to test and validate the results collected by the systems developed as part of the PhD. As a result, you must be capable of passing an ENG1 medical and survival at sea course. You will embrace new challenges and environments and be able to fit into new teams rapidly. You must be able to describe complex issues in a means that is accessible to fishermen with whom you will work. A valid driving license is required.
How to apply
You are requested to send a cover letter stating why you are interested in the PhD, what ideas you could bring to the project, and outline any relevant experience. You are also requested to submit a CV with all qualifications to date. The cover letter and CV should be sent to Prof Michel Kaiser (firstname.lastname@example.org). Candidates are invited to contact Prof Kaiser for an informal discussion about the project.
In addition, to apply you must complete our online application form. Please select PhD programme Marine Biology and include the full project title, reference number and supervisor (Prof MJ Kaiser) on your application form. Ensure that all fields marked as 'required' are complete.
You must complete the section marked project proposal; upload a supporting statement documenting your reasons for applying to this particular PhD project, and why you are an ideal candidate for the position. You will also need to provide a CV, a copy of your degree certificate/s and relevant transcripts. You will be asked to enter details of an academic referee who will be able to provide a technical reference. Until your nominated referee has uploaded their statement, your application will not be marked as complete and will not be considered by the review panel. You must also provide proof of your ability in the English language (if English is not your mother tongue or if you have not already studied for a degree that was taught in English within the last 2 years). We require an IELTS certificate showing an overall score of at least 6.5 with no component scoring less than 6.0 or a TOEFL certificate with a minimum score of 90 points.
Please contact Prof Michel Kaiser (email@example.com) for further information or an informal discussion.
The closing date for applications is 5 March 2021, with interviews held in early March. Applicants must be available to start the PhD as soon as possible, but no later than August 2021.